The difficulties in the development of an expert system for the diagnostics of acute medicinal poisoning

↑

⇐ ПредыдущаяСтр 3 из 5Следующая ⇒

ESTHER – expert system for the diagnostics

of acute drug poisoning

Oleg Larichev, Artyom Asanov, Yevgeny Naryzhny, and Sergey Strahov

Institute for Systems Analysis, Russian Academy of Sciences, 9, pr. 60 let Octjabrja, 117312,

Moscow, Russia

Filatov Children’s Hospital, Moscow, Russia

O. I. Larichev, A. Asanov, Y. Naryzhny, S. Strahov ,ESTHER – Expert System for the Diagnostics of Acute Drug Poisonings. Applications and Innovations in Intelligent Systems IX. Proceedings of the 21 SGES International Conference on Knowledge Based Systems and Applied Artificial Intelligence / Ed. by A.Macintosh, M. Moulton, A. Preece. — Cambridge, UK: Springer-Verlag, 2001.— Pp. 159–168.

Abstract.In recent years, there have been a lot of cases of acute drug intoxication in Russia, and there exist a lack of qualified specialists capable of forming proper diagnosis of intoxication. That is why this research was aimed at developing of an expert system for diagnostics of poisonings caused by overdose or misuse of widespread medicines. Such medicines are available in every family and it is very easy to exceed the critical dosage that could lead to a fatal issue. According to some estimates about a half of all poisonings are provoked by improper use of medicines.

   This paper presents the main concepts of an Expert System for Toxicological Help (ESTHER). The most widespread medicines were combined into 19 groups according to the similarity in poisoning diagnostics and treatment. More than 60 clinical signs used by an expert in diagnostics of intoxications were included. The system was deliberately designed to use only clinical signs of poisonings with the view of using it in ambulances and hospitals of small towns

where accurate laboratory analyses are not available.

The system imitates reasoning of a physician – an expert in toxicology. The ideas of method for knowledge base construction are presented. The architecture of the expert system is discussed in detail as well.

Introduction

In recent years in Russia there was a trend toward fast increasing of a number of poisoning cases for adults as well as for children. This situation makes more urgent and important the problem of poisoning diagnostics and treatment. A considerable change in the acute poisonings pattern has occurred in the last decade. Pharmaceuticals (particularly psychotic drugs), alcohol and surrogates replaced as the main poisonings causes acetic acid and other corrosives, organophosphorous pesticides and other household agents. Approximately one half of all poisoning cases are connected with drug poisoning.

The critical situation with a large number of the intoxications could be significantly softened by the availability of good expertise in hospitals and in ambulance cars. But unfortunately the majority of physicians working in the ambulance service and in hospitals don’t know enough about the reasons of intoxications, their diagnostics and treatment.

Therefore, the important aim of the research is development of an expert system based on the

knowledge of   experienced physicians. Such system could be applied as a useful source of advice for young physicians and physicians of different specialisations.

This paper presents main ideas of ESTHER (Expert System for Toxicological Help). Now the

system is in the process of testing at the Toxicological Centre of Russian Ministry of Health.

From formal point of view, the diagnostic problem could be presented in the following way.

There is the multi-dimensional space of combinations of diagnostic signs values. By 63 diagnostic signs having in the average 3 values on a scale, the number of such combinations equal to 3⁶³. It is necessary to allocate the combinations (clinical situations- CS) into different classes of poisoning. Let us note that such classes could be intersecting ones. Also, it is needed to differentiate for each class the degrees of intoxication: poisoning and severe poisoning, because a course of a patient’s treatment depends on it.

The main difficulties in the application of Artificial Intelligence approach to the solution of this problem are:

1. To be useful an expert system must closely imitate an expert reasoning in the process of

diagnostics for each CS. However, direct presentation to an expert all CS is impossible due to big dimension of the problem.

2. One could not expect receiving from an expert in the explicit form rules used in the process of diagnostics. It is known that expert’s knowledge is predominantly unconscious.

3. A decision for each CS depends on a combination of diagnostic sign values (holistic image of a possible patient) and cannot be imitated by asking an expert to nominate some coefficients of importance for each value of a diagnostic sign.

4.  In any process of expert knowledge elicitation, an expert could make some errors. It is

desirable to have means to discover errors and eliminate them.

Such specific features of the  problem under consideration do not allow to use known approaches [1, 2] and require to find a new approach.

Познавательные статьи:



Как правильно слушать собеседника

Типичные ошибки при выполнении бросков в баскетболе

Принятие христианства на Руси и его значение

Средства массовой информации США